Chapter 7 Matroid Partition Jack Edmonds Introduction by Jack Edmonds This article, “Matroid Partition”, which first appeared in the book edited by George Dantzig and Pete Veinott, is important to me for many reasons: First for per- sonal memories of my mentors, Alan J. Goldman, George Dantzig, and Al Tucker. Second, for memories of close friends, as well as mentors, Al Lehman, Ray Fulker- son, and Alan Hoffman. Third, for memories of Pete Veinott, who, many years after he invited and published the present paper, became a closest friend. And, finally, for memories of how my mixed-blessing obsession with good characterizations and good algorithms developed. Alan Goldman was my boss at the National Bureau of Standards in Washington, D.C., now the National Institutes of Science and Technology, in the suburbs. He meticulously vetted all of my math including this paper, and I would not have been a math researcher at all if he had not encouraged it when I was a university drop-out trying to support a baby and stay-at-home teenage wife. His mentor at Princeton, Al Tucker, through him of course, invited me with my child and wife to be one of the three junior participants in a 1963 Summer of Combinatorics at the Rand Corporation in California, across the road from Muscle Beach. The Bureau chiefs would not approve this so I quit my job at the Bureau so that I could attend. At the end of the summer Alan hired me back with a big raise. Dantzig was and still is the only historically towering person I have known. He cared about me from a few days before my preaching at Rand about blossoms and about good algorithms and good characterizations. There were some eminent com- binatorial hecklers at my presentation but support from Dantzig, and Alan Hoffman, made me brave. Jack Edmonds Department of Combinatorics and Optimization, University of Waterloo, Canada e-mail: [email protected] 199 200 Jack Edmonds I think of Bertrand Russell, Alan Turing, and George Dantzig as the three most important philosophers of the last century. During an infrequent visit to California from Washington, D.C., sometime in the 60s, Dantzig took me, a wife, and three kids, to Marineland and also to see a new shopping mall in order to prove to us that having a ceiling of a certain height in his carefully planned Compact City is as good as a sky. One time when I unexpectedly dropped in on Dantzig, the thrill of my life was him asking me to lecture to his linear programming class about how the number of pivots of a simplex method can grow exponentially for non-degenerate linear programming formulations of shortest path problems, and also asking me to vet contributions for a math programming symposium which he was organizing. One of my great joys with George Dantzig was when a friend working at Hewlett- Packard asked me to come discuss the future of operations research with his artificial intelligence colleagues. I was discouraged when no one I knew in O.R. seemed in- terested in helping—that is, until I asked George. He told my second wife Kathie and me that he was a neighbor and had socialized with Mr. Hewlett, or was it Mr. Packard, for years, and had never been invited to HP, two blocks away. George took over the show and was wonderful. Kathie video-taped it. The next morning he asked if she had made him a copy yet. Al Tucker made me a Research Associate and put me in charge of his Com- binatorics Seminar at Princeton during 1963–64. Combinatorists whom I wanted to meet accepted paying their own way to speak at my ‘Princeton Combinatorics and Games Seminar’. However, except for Ron Graham who came over from Bell, and Moses Richardson who came down from City University, they were unable to schedule their visits. So I hastily organized a Princeton Conference in the spring of 1964 where the eminent seminar invitees could lecture to each other. At that conference I met Al Lehman who led me, by his matroidal treatment of what he called the Shannon switching game, to see that matroids are impor- tant for oracle-based good algorithms and characterizations. I persuaded Al, along with Chris Witzgall, to come work at the Bureau of Standards, and immediately we started looking for people to participate in a two-week Matroid Workshop at the Bu- reau of Standards in autumn 1964. We didn’t find more than six who had even heard of the term ‘matroid’. About twenty serious people came to it, including Ray Fulk- erson, George Minty, Henry Crapo, Dan Younger, Neil Robertson, and Bill Tutte. Within a year it seemed the whole world was discovering matroids. The Bureau was delighted at the prospect of hiring Al Lehman. However, an aftermath of McCartheism left the Bureau with the rule that new employees had to take an oath of loyalty. The early computer-guru, Ida Rhodes, actually tugged at Al’s arm to try to get him to take the oath but he wouldn’t. Instead he took a research job with a Johns Hopkins satellite of the U.S. Army which did not require such an oath. He literally picketed the Matroid Workshop, speaking to whomever would listen about the ‘Bureau of Double Standards’. We stayed friends for the many years until his recent death in Toronto. 7 Matroid Partition 201 At the same workshop, Gian-Carlo Rota conceived of and started organizing the Journal of Combinatorial Theory. He also insisted that the ‘ineffably cacophonic word matroid’ be replaced by ‘combinatorial geometry’. George Minty was an especially sweet and brilliant participant. He wrote a paper which Bob Bland credits with being a precursor of oriented matroids. He spent years afterwards on successfully extending the good algorithm for optimum matchings in a graph to optimum independent sets in a clawfree graph. His work is still the most interesting aspect of matching theory. During the year after the Matroid Workshop, Ray Fulkerson and I regularly spent hours talking math by government telephone between Santa Monica and Washing- ton. Ray and I never did learn how to work computers, and though I think the pro- totype of email did exist back then in our government circles, he and I didn’t know about it. One of the outcomes of our talk was combining a version of the matroid partitioning algorithm described in the paper here with Ray’s interest in doing ev- erything possible by using network flow methods. My huff about him and Ellis Johnson calling the blossom method “a primal- dual method” led me to look for algorithms for network flow problems which were polytime relative to the number of bits in the capacities as well as in the costs. The reason I had presented the blossom method only for 1-matchings is that for b-matchings I could not call it a “good algorithm” until I had figured out how to do that for network flows. Once it’s done for flows, it’s easy to reduce optimum b-matchings to a flow problem and a b-matching problem where the b is ones and twos. Dick Karp was independently developing good algorithms for network flows and so much later I published with Dick instead of, as intended, with Ray and Ellis. I enjoyed working with Ray and I coined the terms “clutter” and “blocker”. I can’t remember who suggested the term “greedy” but it must have been Alan Goldman and probably Ray as well. It was important to me to ask Ray to check with the subadditive set function expert he knew about submodular set functions. When the answer came back that they are probably the same as convex functions of additive set functions, I knew I had a new tiger by the tail. Ray and I liked to show off to each other. I bragged to him about discovering the disjoint branchings theorem, mentioned later. Trouble is, I then became desperate to find quickly a correction of my faulty proof. I think I would have done a better job on the theorem if I had not been frantic to cover my hubris. During a phone call, Ray mentioned that one day later, four months after the Matroid Workshop, there would be a combinatorics workshop in Waterloo. My boss Alan Goldman rescued me as usual and I quickly hopped a plane to Canada to sleep along with George Minty on sofas in Tutte’s living room. Neil Robertson, a meticulous note-taker, had reported to Crispin Nash-Williams on my Matroid Workshop lectures. Crispin, by his own description, was too enthu- siastic about them. He was giving a keynote lecture about matroid partitioning on the first morning of this Waterloo workshop. I felt compelled immediately after his talk to speak for an impromptu hour on the following: 202 Jack Edmonds Theorem 1. A non-negative, monotone, submodular set function, f (S), of the sub- sets S of a finite set E, is called a polymatroid function on E. For any integer- valued polymatroid function on E, let F be the family of subsets J of E such that for every non-empty subset S of J, the cardinality of S is at most f (S). Then M =(E,F) is a matroid. Its rank function is, for every subset A of E, r(A), mean- ing max[cardinality of a subset of A which is in F]=min[ f (S)+cardinality of (A\ S) for any subset S of A].
Details
-
File Typepdf
-
Upload Time-
-
Content LanguagesEnglish
-
Upload UserAnonymous/Not logged-in
-
File Pages20 Page
-
File Size-